Fig 1 - uploaded by Facundo Bre
Content may be subject to copyright.
Artificial neural network architecture (ANN i-h 1-h 2-h n-o). 

Artificial neural network architecture (ANN i-h 1-h 2-h n-o). 

Source publication
Article
Full-text available
Knowing the pressure coefficient on building surfaces is important for the evaluation of wind loads and natural ventilation. The main objective of this paper is to present and to validate a computational modeling approach to accurately predict the mean wind pressure coefficient on the surfaces of flat-, gable- and hip-roofed rectangular buildings....

Contexts in source publication

Context 1
... this work, a feed-forward multilayer ANN is used. Fig. 1 shows the general ANN architecture, which has an input layer, a set of hidden layers and an output layer. In each hidden and output layer, there are artificial neurons interconnected via adaptive weights. These weights are calibrated through a training process with input-output data. For each artificial neuron, there is an activation ...
Context 2
... ANN is a set of unit cells (or artificial neurons) arranged in an input layer, one or more hidden layers, and an output layer. Each neuron is connected to those neurons in the neighboring layers via adaptive weights. Fig. 11 shows the model of a generic neuron j in the hidden layer k, whose output is defined as ...
Context 3
... this work, a feed-forward multilayer ANN is used. Fig. 1 shows the general ANN architecture, which has an input layer, a set of hidden layers and an output layer. In each hidden and output layer, there are artificial neurons interconnected via adaptive weights. These weights are calibrated through a training process with input-output data. For each artificial neuron, there is an activation function, which can be any function with range [−1, 1]; the most common activation functions are the tangent sigmoid and the logarithmic sigmoid ...
Context 4
... ANN is a set of unit cells (or artificial neurons) arranged in an input layer, one or more hidden layers, and an output layer. Each neuron is connected to those neurons in the neighboring layers via adaptive weights. Fig. 11 shows the model of a generic neuron j in the hidden layer k, whose output is defined as ...

Similar publications

Article
Full-text available
The composite tape winding process, which utilizes a tape winding machine and prepreg tapes, provides a promising way to improve the quality of composite products. Nevertheless, the process parameters of composite tape winding have crucial effects on the tensile strength and void content, which are closely related to the performances of the winding...
Article
Full-text available
The paper proposes a prediction methodology for the significant wave height (and implicitly the wave power), based on the artificial neural networks. The proposed approach takes as input data the wind speed values recorded for different time periods. The prediction of significant wave height is useful both for assessment of wave energy as also for...

Citations

... Artificial neural networks (ANN) were the first AI algorithms used for modeling heat pipe systems. An ANN is a computational model capable of simulating the brain's behavior and performing various computational tasks by predicting a number of outputs using a number of inputs [41]. There are various types of neural networks based on different training algorithms, such as the multilayer perceptron neural network (MLPNN), radial basis function (RBF) neural network, convolutional neural network (CNN), etc. ANNs are employed in forecasting, control, modeling, and pattern classification applications [42]. ...
Article
Full-text available
Heat pipe systems have attracted increasing attention recently for application in various heat transfer-involving systems and processes. One of the obstacles in implementing heat pipes in many applications is their difficult-to-model operation due to the many parameters that affect their performance. A promising alternative to classical modeling that emerges to perform accurate modeling of heat pipe systems is artificial intelligence (AI)-based modeling. This research reviews the applications of AI techniques for the modeling and control of heat pipe systems. This work discusses the AI-based modeling of heat pipes focusing on the influence of chosen input parameters and the utilized prediction models in heat pipe applications. The article also highlights various important aspects related to the application of AI models for modeling heat pipe systems, such as the optimal AI model structure, the models overfitting under small datasets conditions, and the use of dimensionless numbers as inputs to the AI models. Also, the application of hybrid AI algorithms (such as metaheuristic optimization algorithms with artificial neural networks) was reviewed and discussed. Next, intelligent control methods for heat pipe systems are investigated and discussed. Finally, future research directions are included for further improving this technology. It was concluded that AI algorithms and models could predict the performance of heat pipe systems accurately and improve their performance substantially.
... ANN [27] is composed of a network of many interconnected neurons or nodes, structured in three main layers as depicted in figure 15: the input layer, hidden layers, and output layer, where the nodes of each layer have no connection between them. Fig. 15 Artificial Neural Network architecture [28] Each interconnected node is a computational unit. Its role then is to make a summation of all the received values x i multiplied with their weight w i , then add a bias b k , which controls the input to the activation function, to the summation of the multiplied values. ...
Preprint
Full-text available
Massive volumes of data generated by large numbers of connected Internet of Things (IoT) devices impact both IoT service and network performance. Packet routing is an ever-important networking feature that enables the transmission of data packets across interconnected IoT networks. Usually, classical proactive and reactive routing protocols use the information of their neighbours to build a global view of the network by the mean of control messages and then calculate a path to the destination. After a packet has been delivered, history is discarded, and only actual information is used to determine new paths. In the present research work, we propose applying a Machine Learning (ML) approach to take advantage of previously successfully taken routing decisions and build new decision models that nodes can use to make intelligent routing decisions. In our proposal, we integrate three machine learning techniques: Decision Trees(DT), Support Vector Machines (SVM), and Artificial Neural Networks (ANN) into the well-known Ad hoc On-demand Distance Vector (AODV) routing protocol and evaluate the efficiency of solution variants for packets routing in IoT-connected smart city scenario. The assessment results show that the proposed ML-based routing solutions outperform the AODV protocol and decrease the use of control messages to more than 50% of the overall communication overhead. Therefore, network performances are more optimized, and network lifetime is also increased.
... where x and y are the input and output of the network, w i , b i and h i are the respective weights, biases and activation functions in the i th layer of the network, i = 1, 2, ..., n, where n is the number of hidden layers in the network. Even though deep neural network has achieved remarkable results by simulating the structure of human brain neural network, the way deep neural networks work [2] is still quite different from human cognition and lack of interpretability, making it difficult to guarantee the credibility of its output. ...
Preprint
Full-text available
Although deep learning has made remarkable progress in processing various types of data such as images, text and speech, they are known to be susceptible to adversarial perturbations: perturbations specifically designed and added to the input to make the target model produce erroneous output. Most of the existing studies on generating adversarial perturbations attempt to perturb the entire input indiscriminately. In this paper, we propose ExploreADV, a general and flexible adversarial attack system that is capable of modeling regional and imperceptible attacks, allowing users to explore various kinds of adversarial examples as needed. We adapt and combine two existing boundary attack methods, DeepFool and Brendel\&Bethge Attack, and propose a mask-constrained adversarial attack system, which generates minimal adversarial perturbations under the pixel-level constraints, namely ``mask-constraints''. We study different ways of generating such mask-constraints considering the variance and importance of the input features, and show that our adversarial attack system offers users good flexibility to focus on sub-regions of inputs, explore imperceptible perturbations and understand the vulnerability of pixels/regions to adversarial attacks. We demonstrate our system to be effective based on extensive experiments and user study.
... YSA sistemlerinde girdi, bilgiyi katmandan geçirir ve çıktı olarak yorumlar. Bu sistemlerin yapısı Şekil 2'de (Bre et at., 2018) gösterilmiştir. Akılcı ilaç kullanımında kullanılan uygulamalar, bazı alt gruplara ayrılarak incelenebilir. ...
... Activation functions define how the weighted sum of the input is transformed into an output, with popular functions, including Rectified Linear Activation (ReLU), Sigmoid, and Tanh, for the hidden layer, and Sigmoid and Softmax for the output layer. In this paper, the output of the transformer (e.g., BERT) encoders are fitted to NN with the following sequence of layers: a layer of 32 neurons, a dropout layer, and an output layer, with Sigmoid as the activation function [34][35][36]. It is worth noting that the NN model was used as a baseline for comparison with the DNN models. ...
Article
Full-text available
Social media platforms such as Twitter are a vital source of information during major events, such as natural disasters. Studies attempting to automatically detect textual communications have mostly focused on machine learning and deep learning algorithms. Recent evidence shows improvement in disaster detection models with the use of contextual word embedding techniques (i.e., transformers) that take the context of a word into consideration, unlike the traditional context-free techniques; however, studies regarding this model are scant. To this end, this paper investigates a selection of ensemble learning models by merging transformers with deep neural network algorithms to assess their performance in detecting informative and non-informative disaster-related Twitter communications. A total of 7613 tweets were used to train and test the models. Results indicate that the ensemble models consistently yield good performance results, with F-score values ranging between 76% and 80%. Simpler transformer variants, such as ELECTRA and Talking-Heads Attention, yielded comparable and superior results compared to the computationally expensive BERT, with F-scores ranging from 80% to 84%, especially when merged with Bi-LSTM. Our findings show that the newer and simpler transformers can be used effectively, with less computational costs, in detecting disaster-related Twitter communications.
... VIV, Galloping), the associated unsteady effect, large-amplitude vibration, surface slanting, or torsional motion are all sources of uncertainty for laser displacement sensors. To this end, many previous efforts have adopted the data-driven approach to solve problems in fluid-structure interaction [15,[36][37][38][39]. Owing to its ability in complex regression, task classification, and superior extrapolation, machine learning (ML) is perhaps the most widely deployed data-driven scheme [40][41][42][43][44][45][46][47]. For example, Snaiki and Wu [48] adopted knowledge-enhanced deep learning to capture complex dynamics and successfully predicted the boundary-layer winds of tropical cyclones. ...
Article
Aerodynamic damping, as the principal determinant of wind-induced vibrations, bears vast significance in building engineering. This study presents a new modular framework, called the Deep Neural Network-Genetic Algorithm (DNN-GA) architecture, to predict aerodynamic damping directly from surface pressure measurement while serving as one of the earliest references for GA applications in aerodynamic damping predictions. The accurate prediction can give building engineers better insights into potential designs' feasibility at substantially reduced costs, benefiting building design and engineering implementations. The DNN-GA is demonstrated on an aeroelastic tapered prism—a nonlinear bi-directional fluid-structure interaction (FSI) system with solid connections to building design—with synchronous, high-fidelity wind tunnel data for training and prediction. With pressure input, the DNN module predicts tip response as the intermediate product, based on which the GA module optimizes for aerodynamic damping in a fully automated workflow. Results showed that the DNN-GA outperformed six benchmark machine learning algorithms by at least 400%. A comparison between the GA module and the traditional Random Decrement Technique (RDT) showed an accuracy improvement of at least 700%. Finally, the DNN-GA predicted the aerodynamic damping for the Vortex-Induced Vibration (VIV), Galloping, and VIV-Galloping regimes with the maximum root-mean-squared and mean-absolute errors of only 3.874×10⁻³ and 3.053×10⁻³, attesting to the method's excellent accuracy and suitability to complex, nonlinear, and many types FSI vibrations. Given its data-driven nature, the DNN-GA is also applicable to experimental, numerical, and even field data, making it an attractive tool for building engineering applications.
... Consider z to be an output of a neuron, and that z is merged with bias b and results of coefficient w with x as an input, and that δ(z) is the output of a sigmoid activation function translated by Bayes, which is formulated as shown in (1) and (2). [9] z= (1) ...
Article
Land use land cover (LULC) usually alludes to the assortment and cataloging of certain activities carried out by humans together with the natural elements on the land. Sentinel satellite images are meant to obtain optical images at high spatial resolution say of about 10m. In this paper, LULC map generation approach using Sentinel satellite images is proposed. Our objective is to classify the entire sentinel image to generate LULC map, which can be further used for predictive analysis. Here, we have used three predominant bands namely NIR, Red and Green to classify the sentinel data with five classes namely Water, Forest, Vegetation, Urban and Open land of silicon city of India. For the proposed dataset, an inclusive exactness of 95% was achieved with neural networks and various deep convolutional neural network architectures.
... The model required half of the effort of wind tunnel tests to predict wind pressure on the roof, even for roof slopes that were not part of the training dataset. More recently, several studies employed novel ANN architectures such as feed-forward multilayer ANN (Bre et al., 2018), and fuzzy neural networks (Fu et al., 2007) to improve accuracy for predicting the wind pressure on buildings. ...
Article
Full-text available
This study used explainable machine learning (XML), a new branch of Machine Learning (ML), to elucidate how ML models make predictions. Three tree-based regression models, Decision Tree (DT), Random Forest (RF), and Extreme Gradient Boost (XGB), were used to predict the normalized mean (C p,mean), fluctuating (C p,rms), minimum (C p,min), and maximum (C p,max) external wind pressure coefficients of a low-rise building with fixed dimensions in urban-like settings for several wind incidence angles. Two types of XML were used-first, an intrinsic explainable method, which relies on the DT structure to explain the inner workings of the model, and second, SHAP (SHapley Additive exPlanations), a post-hoc explanation technique used particularly for the structurally complex XGB. The intrinsic explainable method proved incapable of explaining the deep tree structure of the DT, but SHAP provided valuable insights by revealing various degrees of positive and negative contributions of certain geometric parameters, the wind incidence angle, and the density of buildings that surround a low-rise building. SHAP also illustrated the relationships between the above factors and wind pressure, and its explanations were in line with what is generally accepted in wind engineering, thus confirming the causality of the ML model's predictions.
... Machine Learning (ML) merupakan cabang dari ilmu kecerdasan buatan dimana suatu pembelajaran bisa dilakukan pada sebuah metode sehingga bisa belajar mandiri dalam menyelesaikan suatu problem. ML melakukan proses pada komputer untuk melakukan tindakan mandiri untuk mencapai tingkat akurasi yang tinggi (Alzubi et al., 2018 (Wahyuningrum, 2020), Sedangkan menurut Visualisasi bentuk dari metode NN (Bre et al., 2018) dapat dilihat pada gambar 2 sebagai berikut : ...
Article
Pariwisata merupakan sektor yang sangat penting bagi pemerintah Indonesia, Salah satu bagian dari industri kepariwisataan adalah kebun binatang. Namun dengan semakin banyaknya jumlah kebun binatang di Indonesia pemerintah mengalami kesulitan dalam mengklasifikasikan jenis binatang terutama saat terjadi penambahan hewan baru. Berbagai maca teknologi informasi telah membantu manusia dalam menyelesaikan pekerjaan. Saat ini Machine Learning yang merupakan pendekatan dalam melakukan analisis prediksi terhadap berbagai permasalahan yang dihadapi manusia berkembang dengan pesat. Neural Network merupakan suatu algoritma machine learning yang terbukti tepat dalam memprediksi dan banyak dipergunakan peneliti. Penelitian ini menggunakan Neural Network dalam memprediksi jenis hewan yang ada di kebun binatang sebagai salah satu upaya membantu pemerintah dalam menyediakan informasi yang dibutuhkan. Menggunakan Neural Network diperoleh hasil akurasi sebesar 98% dengan tingkat presisi sebesar 98.1% dan Recall sebesar 98 %.
... The layer that isn't visible analyses the data for the next layer then sends the results to the output layer of the previous layer. Several Machine Learning (ML) and ANN techniques have been introduced to resolve issues on air pollution prediction and forecasting [1]. ANN algorithms are useful in building such models because they can capture the nonlinear behavior of complex atmospheric systems. ...
Article
Full-text available
In recent years, airborne broadcasting has grown more prevalent in cities. Air quality degradation is a severe air pollution issue that exists daily. To forecast the amount of pollutants, Artificial Neural Network (ANN) and Linear Vector Quantization (LVQ) techniques were utilized. The data set dimensions are defined by the pre-processing procedure and the feature extraction mechanism. The ANN model predicts categorization concentration, allowing the LVQ model to classify direct situations with greater accuracy using explanatory factors. The ANN+LVQ model outperformed other technologies in terms of classification accuracy. The raw data was cleaned to improve the accuracy of the prediction algorithms. The pollutants discovered in the collection are NO2, NOx, O3, Benzene, Xylene, NH3, CO, SO2, PM10, NO, and Toluene. The performance of the recommendation and forecast models were tested in this study using two datasets in two distinct experiments. In urban, rural, and industrial settings, the proposed ANN model is successful in detecting air quality and predicting pollution levels. The ANN-LVQ model obtained 90% percent sensitivity, 97.59% accuracy, and 99.46% specificity with 2.43% error rate. The suggested model's accuracy is much greater than that of other current research models.